1,721,062 research outputs found
Going Beyond Counting First Authors in Author Co-citation Analysis
Full text linkThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Parsimonious Online Learning with Kernels and Random Features with Applications to Stochastic Optimal Control
No full textThis thesis explores stochastic optimal control through three interconnected contributions: a theoretical framework, and the development of two regression methods, POLK and POLRF, tailored to high-dimensional regression challenges in this domain.
In the first part, we propose a general framework for solving stochastic optimal control problems using dynamic programming. A key contribution is the decomposition of total error at each time step into local approximation errors and propagated errors from future steps. This decomposition provides new theoretical insights into error backpropagation, a relatively underexplored area in stochastic control. While this framework establishes a solid foundation, its applicability is limited to problems amenable to dynamic programming.
The second part introduces POLK (Parsimonious Online Learning with Kernels), a kernel-based regression method enhanced with sparsification to control model complexity. POLK achieves efficient evaluation times and performs well on intricate regression tasks with localised variations. We further improve its empirical convergence speed by incorporating a second gradient term while maintaining theoretical guarantees. However, the computational cost of the sparsification process restricts its scalability to larger datasets, limiting its practical use in high-dimensional settings.
In the third part, we develop POLRF (Parsimonious Online Learning with Random Features), a scalable alternative to kernel-based methods. POLRF replaces kernels with random features, dynamically adapting its feature map distribution through sparsification. This adaptation resembles kernel learning, allowing POLRF to construct problem-specific representations. We establish a theoretical framework for POLRF, including convergence guarantees in a trajectory-dependent reproducing kernel Hilbert space (RKHS). Empirical results on the max-call option pricing problem demonstrate POLRF's competitive performance, including faster convergence and robustness in high dimensions. Despite its promise, open questions remain regarding the properties of the constructed RKHS and the optimisation of random feature distributions for specific problems.CS
Estimating the Term Structure of Interest Rates Using Linear Programming and Parsimonious Functional Forms.
No full textVariations on the Author
Full text link“Variations on the Author” discusses two of Eduardo Coutinho’s recent films (Um Dia na Vida, from 2010, and Últimas Conversas, posthumously released in 2015) and their contribution to the general question of documentary authorship. The director’s filmography is characterized by a consistent yet self-effacing form of authorial self-inscription: Coutinho often features as an interviewer that rather than express opinions propels discourses; an interviewer that is good at listening. This mode of self-inscription characterizes him as an author who is not expressive but who is nonetheless markedly present on the screen. In Um Dia na Vida, however, Coutinho is completely absent form the image, while Últimas Conversas, on the contrary, includes a confessional prologue that moves the director from the margins to the center of his films. This article examines the ways in which these works stand out in the filmography of a director who offers new insights into the notion of cinematic authorship
Closed form approximation methods for portfolio valuation and risk management
No full textIn this thesis we present three closed form approximation methods for portfolio valuation and risk management.
The first chapter is titled ``Kernel methods for portfolio valuation and risk management'', and is a joint work with Damir Filipovi\'c (SFI and EPFL). We introduce a simulation method for portfolio valuation and risk management building on machine learning with kernels. We estimate the value process of a portfolio from a finite sample of its cumulative cash flow. The estimator of a portfolio value process is given in closed form thanks to a suitable choice of the kernel. We show asymptotic consistency and derive finite sample error bounds under conditions that are suitable for finance applications. Numerical experiments show good results for examples in dimensions 12 and 36.
The second chapter is titled ``Ensemble learning for portfolio valuation and risk management'', and is a joint work with Damir Filipovi\'c (SFI and EPFL). We introduce a second closed form estimator of the value process of a portfolio. This estimator is based on ensemble learning methods with regression trees. In contrast to the first estimator, this estimator is fast to construct, and readily scalable with sample size and path space dimension. We also show how this estimator can be applied to derive a closed form estimator of the value process of a Bermudan option. Numerical experiments show good results for examples in dimensions 12 and 36.
The third chapter is titled ``Polynomial approximation for interest rate derivatives valuation'', and is a joint work with Damir Filipovi\'c (SFI and EPFL). We model the risk-neutral discount factor using a linear-rational model with a polynomial diffusion. Thanks to this modelling, the zero-coupon bond price, the continuously compounded overnight rate, and the forward rate are given in closed form. For an interest rate derivative whose price is not given in closed form, such as caplets, futures, and futures options, we use Bernstein polynomials to derive a closed form approximation of such a price that satisfies an -error bound.CS
Collateralized Debt Obligations: An Examination of the Various Structures, The Market Place, Credit Risk, and Over-Collateralization.
No full textAppropriate Similarity Measures for Author Cocitation Analysis
Full text linkWe provide a number of new insights into the methodological discussion about author cocitation analysis. We first argue that the use of the Pearson correlation for measuring the similarity between authors’ cocitation profiles is not very satisfactory. We then discuss what kind of similarity measures may be used as an alternative to the Pearson correlation. We consider three similarity measures in particular. One is the well-known cosine. The other two similarity measures have not been used before in the bibliometric literature. Finally, we show by means of an example that our findings have a high practical relevance.information science;Pearson correlation;cosine;similarity measure;author cocitation analysis
Financial Risk Management with Machine Learning
No full textThis thesis consists of three applications of machine learning techniques to risk management. The first chapter proposes a deep learning approach to estimate physical forward default intensities of companies. Default probabilities are computed using artificial neural networks to estimate the intensities of the inhomogeneous Poisson processes governing default process. The major contribution to previous literature is to allow the estimation of non-linear forward intensities by using neural networks instead of classical maximum likelihood estimation. The model specification allows an easy replication of previous literature using linear assumption and shows the improvement that can be achieved. The second chapter, titled `Causal Networks with Neural Networks` is a co-authored work with Damir Filipovic (SFI & EPFL), Negar Kiyavash (EPFL) and Jalal Etesami (EPFL). We develop a data-driven framework to identify the interconnections between firms using an information-theoretic measure. This measure generalizes Granger causality and is capable of detecting nonlinear relationships within a network. Moreover, we develop an algorithm using recurrent neural networks and Granger causality to identify the interconnections of high-dimensional nonlinear systems. The outcome of this algorithm is the causal graph encoding the interconnections among the firms. These causal graphs can be used as preliminary feature selection for another predictive model or for systemic risk management. We evaluate the performance of our algorithm using both synthetic linear and nonlinear experiments and apply it to the daily stock returns of US listed firms and infer their interconnections from 1990 to 2020. The third chapter, titled `StockTwits Classified Sentiment and Stock Returns` is a co-authored work with Damir Filipovic (SFI & EPFL). We classify the sentiment of a large sample of StockTwits messages as bullish, bearish or neutral, and create a stock-aggregate daily sentiment polarity measure. Polarity is positively associated with contemporaneous stock returns. On average, polarity is not able to predict next-day stock returns. But when we focus on specific events, defined as sudden peaks of message volume, polarity has predictive power on abnormal returns. Polarity-sorted portfolios illustrate the economic relevance of our sentiment measure.CS
Function Learning with Financial Applications
No full textThis thesis sets itself the goal of investigating function learning approaches. The topic is extremely relevant today, since the abundance of data, jointly with the technological progress which has enhanced computing power, now allow to represent functions of interest with unprecedented prowess starting from available observations. Finance and financial engineering undoubtedly have compelling reasons to exploit such a trend, namely the need to perform computations fast and accurately. Since financial models rely on stochasticity and on complex numerical solutions, it becomes vital both in the academia and for practitioners to deploy methods which overcome the burdens of existing ones.
Many issues still remain open. Interpretability and lack of theoretical understanding of existing methods, such as neural networks; data scarcity in some of the financial applications; large data which become intractable and the related curse of dimensionality; the handling of time series and computational issues are among the aspects which deserve further investigation and which will make the object of this work.CS
- …
